fe-safe™ from Safe Technology

John Draper

fe-safe: what we do

fe-safe

The leading suite of software for fatigue design

Sold world wide to companies that design everything from

mobile phones to heavy engineering structures,

heart valves to engines

nuclear plant to wind turbines

fe-safe: what we do

fe-safe

The leading suite of software for fatigue design

Sold world wide to companies that design everything from

mobile phones to heavy engineering structures,

heart valves to engines

nuclear plant to wind turbines

High temperature fatigue of engines

High temperature fatigue of enginesFatigue of welded joints - VerityTM

Developed and patented by the Battelle

Institute. Licensed to Safe Technology

- world-wide agreement in place

- we take 40% of the software revenue

A large Joint Industry Panel (JIP) has guided

the research and validated the method.

The developer has received many awards -

Society of Automotive Engineers Henry Ford II

medal

Time Magazine 2005 maths innovator.

The potential market for the VerityTM method

is vast - for example -Very small components - drive shafts, bicycle

parts

Exhaust systems

Soldered joints and printed

circuit boards

High temperature fatigue of enginesFatigue of welded joints - VerityTM

Developed and patented by the Battelle

Institute. Licensed to Safe Technology

- world-wide agreement in place

- we take 40% of the software revenue

A large Joint Industry Panel (JIP) has guided

the research and validated the method.

The developer has received many awards -

Society of Automotive Engineers Henry Ford II

medal

Time Magazine 2005 maths innovator.

The potential market for the VerityTM method

is vast - for example -Very small components - drive shafts, bicycle

parts

Exhaust systems

Soldered joints and printed

circuit boards

Car bodies and car assembly plant

“This will save millions of dollars” - Ford

Fatigue of welded joints - VerityTM

Car bodies and car assembly plant

“This will save millions of dollars” - Ford

Fatigue of trucks and off-highway vehicles

Fatigue of pressure vessels

Fatigue of pressure vesselsFatigue of medical equipment

Medical equipment -

X-ray treatment and mobile

MRI scanners

Arterial stents

Fatigue of medical equipment

Medical equipment -

X-ray treatment and mobile

MRI scanners

Arterial stents

Fatigue of welded joints - VerityTMFatigue of medical equipment

Pressure vesselsMedical equipment -

X-ray treatment and mobile

MRI scanners

Arterial stents

Fatigue of welded joints - VerityTM…and fatigue of….

Pressure vesselsMedical equipment -

X-ray treatment and

mobile MRI scanners

Arterial stents

Safe Technology Limited

• Headquarters in Sheffield UK

• Safe Technology US is a wholly-owned subsidiary based in

Michigan with an office in Los Angeles

• World-wide distributor network includes ANSYS re-sellers

• fe-safe is sold worldwide by Dassault Systemes Simulia Corp

• Revenue growth is typically 30% per year

12

INTERFACES

Input/output

ABAQUS .fil

ABAQUS .odb

NASTRAN f06

NASTRAN op2

ANSYS .rst

I-DEAS .unv

Pro/M s01..., d01

General .csv

Output

Hypermesh .hmres

PATRAN

FEMVIEW

CADFIX

FEMAP

Redesign

Design

FEAABAQUS, ANSYS

I-DEAS,

NASTRAN, Pro/E

Stress

resultsfe-safe

fefe-safe durability analysis from FEA

Loading

Lifecontours

Mean stress

Stress

amplitude

Haigh Diagram

Stress

amplitude

N cycles

S-N curve

ampP (sf-sm)/E 2.00E+05 ampE ampP amp amp/Sao

0.000328 0.001855 1.25113 0.006059 0.313619 0.0019 0.000947 0.002848 1.194403

0.000328 0.001793 1.209275 0.005813 0.313619 0.001823 0.000947 0.00277 1.162002

0.000328 0.001731 1.16742 0.005567 0.313619 0.001746 0.000947 0.002693 1.129602

0.000328 0.001669 1.125565 0.00532 0.313619 0.001669 0.000947 0.002616 1.097201

0.000328 0.001607 1.08371 0.005074 0.313619 0.001591 0.000947 0.002539 1.064801

0.000328 0.001545 1.041855 0.004828 0.313619 0.001514 0.000947 0.002461 1.0324

0.000328 0.001483 1 0.004581 0.313619 0.001437 0.000947 0.002384 1

0.000328 0.001421 0.958145 0.004335 0.313619 0.00136 0.000947 0.002307 0.9676

0.000328 0.001359 0.91629 0.004089 0.313619 0.001282 0.000947 0.00223 0.935199

0.000328 0.001297 0.874435 0.003842 0.313619 0.001205 0.000947 0.002152 0.902799

0.000328 0.001235 0.83258 0.003596 0.313619 0.001128 0.000947 0.002075 0.870398

0.000328 0.001173 0.790725 0.00335 0.313619 0.001051 0.000947 0.001998 0.837998

0.000328 0.001111 0.74887 0.003103 0.313619 0.000973 0.000947 0.001921 0.805597

0.000328 0.001048 0.707015 0.002857 0.313619 0.000896 0.000947 0.001843 0.773197

0.000328 0.000986 0.665161 0.002611 0.313619 0.000819 0.000947 0.001766 0.740796

Location of max stress.

Calculate stress amplitude and

mean stress

Durability by design?

Dana Automotive Systems Group Case Study

Part of an automotive driveshaft assembly joint.

Cracks did not start from the maximum stress location.

This was corroborated by lab tests on actual specimens

fe-safe life contours Stress contours

Max principal stress

Shortest life

De

elastic-plasticstrain

Fatigue calculations using local strains

Same stress, same life

For crack initiation, life is determined by surface stresses and strains

Distance, r

rc

Whether the crack will propagate or not is defined by the

stress some distance below the surface.

This „critical distance‟ is a material property obtained from

DKth

“Notch sensitivity”

This stress determines if the crack will initiate

This stress determines if the crack will

propagate to failure

Will the crack grow ? - Critical distance methods

s

Distance, r

rc

Whether the crack will propagate or not is defined by the

stress some distance below the surface.

This „critical distance‟ is a material property

“Notch sensitivity”

This stress determines if the crack will initiate

This stress determines if the crack will

propagate to failure

High strength steel – critical distance is small (say 0.1mm)

Very „notch sensitive‟.

Cracks will usually propagate to failure

Critical distance methods

s

Distance, r

“Notch sensitivity”

This stress determines if the crack will initiate

This stress determines if the crack will

propagate to failure

High strength steel – critical distance is small (say 0.1mm)

Very „notch sensitive‟.

Cracks will usually propagate to failure

Grey iron – critical distance is large (say 2 mm)

Very „notch insensitive‟.

Cracks will often not propagate to failure

Critical distance methods

s

So now we can include…..the stress gradient „size effect‟ in fatigue

Shorter life to failure

…but same life to crack initiation.

So now we can include…..the difference between axial and bending

Axial loading gives

shorter life to failure

…but same life to crack initiation.

HF Moore - 1927

Crack initiation fatigue lives were related to the amount of

plasticity in notches

Plasticity

What do we mean by „plasticity‟ ?

Plasticity

Modern fatigue analysis is the study of the effects of small amounts of

inelasticity

Battelle 1947. Fatigue manual for US Navy

0

50

100

150

200

250

300

350

400

0 0.002 0.004 0.006 0.008

Strain

Str

ess:

MP

a0

50

100

150

200

250

300

350

400

0 0.002 0.004 0.006 0.008

Strain

Str

ess:

MP

a

SAE1030 steel

‘Yield stress’ =

0.2% proof stress

from tensile test

‘Yield stress’ =

0.2% proof stress

from tensile test

Cyclic test

Onset of plasticity

in fatigue

“we should look at that portion of the

stress-strain diagram between the

beginning of the tiniest permanent

deformation and the yield strength”

Plasticity

Fatigue analysis methods

What causes fatigue cracking ?

Plasticity is necessary to initiate fatigue cracks.

Tresca and von Mises are methods of calculating when a material

will „yield‟.

Tresca and von Mises are both based on shear stresses.

Therefore – shear stresses initiate fatigue cracks

(this theory is from around 1900)

Findley, 1959

Biaxial fatigue

Findley, 1959 - combine shear and normal stresses

Biaxial fatigue

Principal stresses may change orientation

Biaxial fatigue

Biaxial fatigue

Principal stresses may change orientation

Biaxial fatigue

Principal stresses may change orientation

Biaxial fatigue

Critical plane analysis searches for the most damaged plane

Fatigue analysis methods

1. If we use elastic FEA, we need a fatigue-based

plasticity correction to calculate elastic-plastic

stress and strain.

765

Life (cycles)

Str

ess m

ax/m

in (

MP

a)

-100

0

100

200

300

400

101010

SAE10302. Calculate combinations of shear stress-strain and

normal stress-strain.

3. Perform critical plane searching to find the

direction of crack initiation.

Metal fatigue - 1927

The endurance limit amplitude for a mixture of large and small

cycles is much lower than the constant amplitude endurance

limit.

Fatigue analysis methods

e

Constant amplitude

endurance limit

De

2

Endurance 2Nf

e

damaging

Endurance 2Nf

De

2

e

damaging less damaging

Endurance 2Nf

De

2

e

damaging less damaging non- damaging

Endurance 2Nf

De

2

Fatigue analysis methods

1. Calculate elastic-plastic stress and strain from elastic

FEA stresses.

2. Calculate combinations of shear stress-strain and

normal stress-strain.

3. Perform critical plane searching to find the direction

of crack initiation.

4. Modify the endurance limit to allow for the interaction

between large and small cycles.

5. If we are prepared to accept cracks, we use „critical

distance‟ methods to increase the allowable stresses.

765

Life (cycles)

Str

ess m

ax/m

in (

MP

a)

-100

0

100

200

300

400

101010

SAE1030

rc

Dso

rc

Dso

Local strain analysis allows for changes in mean stress caused by local plasticity

Elastic strains „pull‟ notch strains close to zero when load is removed

Tensile residual stress

Biaxial fatigue

(2 ) (2 )2

f

f f fb cN N

E

sDe e

Strain-life for uniaxial stress

max 1.65 (2 ) 1.75 (2 )2 2

fNf f f

b cN NE

sD Dee

Kandil-Brown-Miller „shear + normal‟

Combined shear and

normal strainUses uniaxial materials properties so

no extra materials data is needed.

1.0E-04

1.0E-03

1.0E-02

1.0E-01

1.0E+00

1.00E+00 1.00E+01 1.00E+02 1.00E+03 1.00E+04 1.00E+05 1.00E+06 1.00E+07 1.00E+08

Endurance 2Nf

2

eD

0

100

200

300

400

500

600

0 0.005 0.01 0.015 0.02 0.025 0.03

Strain

Str

ess (

MP

a)

Cyclic stress-strain curve

Used for calculating mean stresses. Used

for converting elastic FEA stresses into

elastic-plastic stress and strain.

Corrected for the effect of biaxial stresses.

Elastic-plastic strain-life curve

Used for calculating cycle-by-cycle fatigue

damage.

Corrected for the effect of biaxial stresses

to give a Brown-Miller analysis.

Materials data – effect of temperature

1.0E-04

1.0E-03

1.0E-02

1.0E-01

1.0E+00

1.00E+00 1.00E+01 1.00E+02 1.00E+03 1.00E+04 1.00E+05 1.00E+06 1.00E+07 1.00E+08

Endurance 2Nf

2

eD

0

100

200

300

400

500

600

0 0.005 0.01 0.015 0.02 0.025 0.03

Strain

Str

ess (

MP

a)

Cyclic stress-strain curve

Used for calculating mean stresses. Used

for converting elastic FEA stresses into

elastic-plastic stress and strain.

Corrected for the effect of biaxial stresses.

Elastic-plastic strain-life curve

Used for calculating cycle-by-cycle fatigue

damage.

Corrected for the effect of biaxial stresses

to give a Brown-Miller analysis.

Materials data – effect of temperature

Other environmental effects may include strain rate effects, strain ageing,

oxidation, stress relaxation and creep.

Forming and assembly stresses

Sheet Metal Forming Simulation

PunchBlank Holder

Die

Forming and assembly stresses

Fatigue life contours for a Ford oil-pan

(a) excluding and (b) including effects of

forming process

(a) (b)

Forming and assembly stresses

Assembly stresses, residual stresses etc, can be included.

Mean stress

Stress

amplitude

Haigh Diagram

Stress

amplitude

N cycles

S-N curve

ampP (sf-sm)/E 2.00E+05 ampE ampP amp amp/Sao

0.000328 0.001855 1.25113 0.006059 0.313619 0.0019 0.000947 0.002848 1.194403

0.000328 0.001793 1.209275 0.005813 0.313619 0.001823 0.000947 0.00277 1.162002

0.000328 0.001731 1.16742 0.005567 0.313619 0.001746 0.000947 0.002693 1.129602

0.000328 0.001669 1.125565 0.00532 0.313619 0.001669 0.000947 0.002616 1.097201

0.000328 0.001607 1.08371 0.005074 0.313619 0.001591 0.000947 0.002539 1.064801

0.000328 0.001545 1.041855 0.004828 0.313619 0.001514 0.000947 0.002461 1.0324

0.000328 0.001483 1 0.004581 0.313619 0.001437 0.000947 0.002384 1

0.000328 0.001421 0.958145 0.004335 0.313619 0.00136 0.000947 0.002307 0.9676

0.000328 0.001359 0.91629 0.004089 0.313619 0.001282 0.000947 0.00223 0.935199

0.000328 0.001297 0.874435 0.003842 0.313619 0.001205 0.000947 0.002152 0.902799

0.000328 0.001235 0.83258 0.003596 0.313619 0.001128 0.000947 0.002075 0.870398

0.000328 0.001173 0.790725 0.00335 0.313619 0.001051 0.000947 0.001998 0.837998

0.000328 0.001111 0.74887 0.003103 0.313619 0.000973 0.000947 0.001921 0.805597

0.000328 0.001048 0.707015 0.002857 0.313619 0.000896 0.000947 0.001843 0.773197

0.000328 0.000986 0.665161 0.002611 0.313619 0.000819 0.000947 0.001766 0.740796

Location of max stress.

Calculate stress amplitude and

mean stress

Durability by design?

Duty cycle

Stress

Material

data

Fatigue

analysisLife

Durability by design

All stresses and

temperatures

Define failure criteria –

based on risk

Duty cycles

Not just the

„most severe‟

Life

contours

1.0E-04

1.0E-03

1.0E-02

1.0E-01

1.0E+00

1.00E+00 1.00E+01 1.00E+02 1.00E+03 1.00E+04 1.00E+05 1.00E+06 1.00E+07 1.00E+08

Endurance 2Nf

2

eD

1.0E-04

1.0E-03

1.0E-02

1.0E-01

1.0E+00

1.00E+00 1.00E+01 1.00E+02 1.00E+03 1.00E+04 1.00E+05 1.00E+06 1.00E+07 1.00E+08

Endurance 2Nf

1.0E-04

1.0E-03

1.0E-02

1.0E-01

1.0E+00

1.00E+00 1.00E+01 1.00E+02 1.00E+03 1.00E+04 1.00E+05 1.00E+06 1.00E+07 1.00E+08

Endurance 2Nf

2

eD

0

100

200

300

400

500

600

0 0.005 0.01 0.015 0.02 0.025 0.03

Strain

Str

ess (

MP

a)

Strain-based,

high temp.

Biaxial, strain-based,

critical plane, critical

distance

INTERFACES

Input/output

ABAQUS .fil

ABAQUS .odb

NASTRAN f06

NASTRAN op2

ANSYS .rst

I-DEAS .unv

Pro/M s01..., d01

General .csv

Output

Hypermesh .hmres

PATRAN

FEMVIEW

CADFIX

FEMAP

fefe-safe durability analysis from FEA

Redesign

Design

FEAABAQUS, ANSYS

I-DEAS,

NASTRAN, Pro/E

Stress

resultsfe-safe

Loading

Lifecontours

Modern fatigue analysis is the study of the effects of small amounts of

inelasticity

Battelle 1947. Fatigue manual for US Navy

0

50

100

150

200

250

300

350

400

0 0.002 0.004 0.006 0.008

Strain

Str

ess:

MP

a0

50

100

150

200

250

300

350

400

0 0.002 0.004 0.006 0.008

Strain

Str

ess:

MP

a

SAE1030 steel

‘Yield stress’ =

0.2% proof stress

from tensile test

‘Yield stress’ =

0.2% proof stress

from tensile test

Cyclic test

Onset of plasticity

in fatigue

“we should look at that portion of the

stress-strain diagram between the

beginning of the tiniest permanent

deformation and the yield strength”

Plasticity

s

e

Elastic FEA

Amplitude is smaller

Mean stress is lower

s

e

S-N curves with elastic FEA

765

Life (cycles)

Str

es

s m

ax

/min

(M

Pa

)

Cycles at R=0. Elastic FEA (grey) and actual (black)

-100

0

100

200

300

400

101010

20

14

-

T6Strain

Str

ess

as

eas

as

2014-T6

S-N curves with elastic FEA

765

Life (cycles)

Str

es

s m

ax

/min

(M

Pa

)

Cycles at R=0. Elastic FEA (grey) and actual (black)

-100

0

100

200

300

400

101010

SAE1030

20

14

-

T6Strain

Str

ess

as

eas

asS-N curves with elastic FEA

Even for high cycle fatigue, mean stresses will be wrong unless we do a plasticity correction.

Without a plasticity correction, we are limited to „infinite life‟ design at mean stresses close to zero.

765

Life (cycles)

Str

ess m

ax/m

in (

MP

a)

-100

0

100

200

300

400

101010

SAE1030

S-N curves with elastic FEA

Changes in mean stress caused by local plasticity

S-N curves with elastic FEA

S-N curves with elastic FEA

Strain

Str

es

s

ms

ems

as

eas

as

Plasticity reduces the mean stress

We may try to adjust the Goodman diagram to

compensate for the error in calculating mean

stresses

S-N curves with elastic FEA

Sa

Sm

00 UTS

local

notched

local - elastic

Strain

Str

es

s

ms

ems

as

eas

as

SS

tK

S-N curves with elastic FEA

0

50

100

150

200

0 100 200 300 400 500 600

Mean stress (MPa)

Str

ess a

mp

litu

de M

Pa Local ,e e

a ms s

Notched ,a mS S

Local ,a ms s

‘Notched – brittle’

‘Notched – ductile’

MANTEN steel

SS

tK

S-N curves with elastic FEA

0

50

100

150

0 100 200 300 400 500 600 700

Mean stress (MPa)

Str

ess a

mp

litu

de M

Pa

Local ,e ea ms s

Notched ,a mS S

Local ,a ms s

‘Notched – brittle’

‘Notched – ductile’

SS

tK

SAE1045 steel

S-N curves with elastic FEA

0

50

100

150

200

0 100 200 300 400 500 600

Mean stress (MPa)

Str

ess a

mp

litu

de M

Pa Local ,e e

a ms s

Notched ,a mS S

Local ,a ms s

‘Notched – brittle’

‘Notched – ductile’

The FKM mean stress correction diagram is an empirical estimate of this

plasticity effect

But the for complex loading the mean stresses will be wrong because

plasticity is being ignored.

FKM Guidelines Figure 4.3.3

S-N curves with elastic FEA

The FKM mean stress correction diagram is an empirical estimate of this

plasticity effect

But the for complex loading the mean stresses will be wrong because

plasticity is being ignored.

FKM Guidelines Figure 4.3.3

S-N curves with elastic FEA

• there is some yielding, even for high cycle fatigue

• there is even more yielding if the mean stresses are not zero

• yielding changes the mean stresses from cycle to cycle

Therefore the statement …...

“for high cycle fatigue there is no difference between the S-N curve and strain-life

approaches to fatigue” …...is not true

An essential feature of fatigue analysis from elastic FEA is the plasticity correction

S-N curves with elastic FEA

RESULTS

• fatigue lives and crack sites

• how much the stresses must be changed to

achieve the design life

• probability of failure at design life

• probability of survival at specified lives -

to predict warranty claims 99.7

99.8

99.9

100

1 10 100 1000 10000 100000 1000000

Miles

Su

rviv

al (%

)

User profile 1

User profile 2

• which loads need to be included during lab testing

• fatigue lives and crack sites

• how much the stresses must be changed to

achieve the design life

• probability of failure at design life

• probability of survival at specified lives -

to predict warranty claims 99.7

99.8

99.9

100

1 10 100 1000 10000 100000 1000000

Miles

Su

rviv

al (%

)

User profile 1

User profile 2

• which loads need to be included during lab testing

RESULTS

fe-safe has two methods

FRF - fatigue reserve factor

From a standard or user-defined mean stress

curve.

Applies only to infinite life design.

FOS – factor of strength for specified life or lives

An iterative process -

Scales the elastic FEA stresses

Recalculates the plasticity for the whole stress history

Recalculates the life

Repeats until it finds the scale factor to give the required life

Applies for finite and infinite life

Stress amplitude

Mean stress

RESULTS

• fatigue lives and crack sites

• how much the stresses must be changed to

achieve the design life

• probability of failure at design life

• probability of survival at specified lives -

to predict warranty claims 99.7

99.8

99.9

100

1 10 100 1000 10000 100000 1000000

Miles

Su

rviv

al (%

)

User profile 1

User profile 2

• which loads need to be included during lab testing

fe-safe combines material and load variability

Probability of survival

Uses Weibull distribution of

fatigue strength and Gaussian

variability in load values

Endurance

De

2

Loading

• Dang Van diagram

• max stress at each node

• max stress / yield stress

• max stress / UTS

• Haigh diagram, Smith diagram

• Time histories of stress tensor, principal stress/strain ...

RESULTS

Examples of fe-safe analysis

3 unit load linear elastic

FE analyses

Forged aluminium alloy suspension component

fatigue life contoursTest life : 41000 miles (long cracks)

fe-safe : 27000 miles (crack initiation)

Forged aluminium alloy suspension component

BEARING STEEL CHARACHTERISTICS

MECHANICAL

• High Hardness (HRC 58-65)

• Very High Tensile Strength (>300 ksi)

• Excellent Wear Resistance

• Excellent Rolling Contact Fatigue Endurance

Bearing-grade steel fatigue

FE METHODS

TESTING

METALLURGIC STUDY

ANALYTICAL FATIGUE ASSESSMENT

FKM-Guideline fe-safe

Bearing-grade steel fatigue

fe-safe detects contact and uses more complex analysis methods

Bearing-grade steel fatigue

Inputs for FEA/fe-safeTM analysis

• Life Theory

Brown-Miller strain-life criteria used. Trials made with

other criteria, such as Principal Stress, BM Combined,

Mises, and Principal Strain.

• Surface Effect

Default method in fe-safeTM worked well with Ra

surface finish as manufactured on the real parts.

• Material Strain-Life Properties

SAE52100 in fe-safeTM database utilized with some

modification per INA USA 52100 test data.

Tensile Test Sample with Bearing Steel

Bearing-grade steel fatigue

fe-safeTM result

Results: Weibull plot of real test data at

20KN load. L50 (test) ≈ 110K cycles.

From FEA/fe-safeTM with Brown-Miller

strain-life criteria and Morrow mean

stress correction, L50 (analysis) ≈ 92K

cycles.

Analysis results well within Weibull

confidence bands at L50.

Note wide range of real results, inherent

to bearing grade steel. (52100)

Weibull Plot of Tripod Roller Test at 20KN

Bearing-grade steel fatigue

Results: Weibull plot of real test data at

21KN load. L50 (test) ≈ 55K cycles.

From FEA/fe-safeTM with Brown-Miller

strain-life criteria and Morrow mean stress

correction, L50 (analysis) ≈ 58K cycles.

Analysis results well within Weibull

confidence bands at L50.

Note wide range of real results, inherent to

bearing grade steel. (52100)

Weibull Plot of Tripod Roller Test at 21KN

fe-safeTM result

Bearing-grade steel fatigue

Comparison of fe-safe and the „S-N + Haigh‟ method

„S-N +Haigh‟ fe-safe

Actual crack site

Fatigue life contours

fe-safe™ User‟s Group Meeting

October 14, 2008

Slide 93

• ASME B&PV (Section VIII div. 2) code design analysis procedures date to the 1960‟s

– Based on plate and shell theory (2D elastic)

– Stresses are decomposed into various classes (primary, secondary, local, thermal, membrane, bending)

– Stresses resolved on Stress Classification Lines (SCL) or “cut-lines” by linearization (analyst chooses cut-lines)

• Choosing SCL‟s and classifying stresses can be difficult and time consuming

Comparison of fe-safe and ASME B&PV

fe-safe™ User‟s Group Meeting

October 14, 2008

Slide 94

Piping Test Section

Comparison of fe-safe and ASME B&PV

fe-safe™ User‟s Group Meeting

October 14, 2008

Slide 95

0

100

200

300

400

500

600

0 20 40 60 80 100 120 140 160 180 200

Time (seconds)

Bu

lk W

ate

r T

em

pe

ratu

re (

F)

0

50

100

150

200

250

300

Test

Secti

on

Flo

w R

ate

(g

pm

)

Pre

ssu

re*1

0

(psi)

Temperature

Pressure

Flow Rate

Comparison of fe-safe and ASME B&PV

Temperature, Pressure and Flow Rate Time Histories

fe-safe™ User‟s Group Meeting

October 14, 2008

Slide 96

Cracked Spacer Ring

Comparison of fe-safe and ASME B&PV

fe-safe™ User‟s Group Meeting

October 14, 2008

Slide 97

Observations

fe-safe™ compares well with the ASME B&PV code fatigue

procedure without need of SCL‟s

…. it just works automatically

Comparison of fe-safe and ASME B&PV

Mike Otto

October 14, 2008

Fatigue Life of a Supercharger Torsion Isolator Spring

One Option: Single Spring Isolator (SSI)

SSI comprised of three “simple” parts.

Hubs are press fit to the drive shafts. The spring is captured within an internal pocket.

The spring winds and unwinds to “absorb” engine speed variation.

Spring rate is selected to isolate the supercharger from the engine‟s torsional vibrations.

Gear

Rear Hub

Spring

Front Hub

SSI comprised of three “simple” parts.

Hubs are press fit to the drive shafts. The spring is captured within an internal pocket.

The spring winds and unwinds to “absorb” engine speed variation.

Spring rate is selected to isolate the supercharger from the engine‟s torsional vibrations.

Gear

Rear Hub

Spring

Front Hub

Fatigue Life of a Supercharger Torsion Isolator Spring

Vehicle Load Signature

• The torque signature represents engine acceleration.

• High sampling frequency is required to capture “peaks” in signal.

“As Measured” Torque Signature

Fatigue Life of a Supercharger Torsion Isolator Spring

fe-safe life contours

Residual Stress: +103 MPa; fe-safe life = 67.5K cycles

Single part failed at

60K cycles.

Crack

developed

here.

Fatigue Life of a Supercharger Torsion Isolator Spring

Problem: Crack at root of

valve pocket radius

“ABAQUS and fe-safe

used to accurately predict

location and time to crack

initiation”

fe-safe fatigue contours

Crack initiation site

Federal Mogul - Modern diesel piston

Allows analysis of the surface nodes – for faster analysis

Allows stress gradient/critical distance correction from surface nodes

Allow faster fatigue solvers for surface nodes

Inner and outer surfaces indentified

fe-safe - identification of surfaces

Identifies all surfaces in a model or assembly

User can define hot-spot criteria – e.g. FOS < 1.2

„What –if‟ investigations can be applied to hotspots

Stress gradient/critical distance corrections can be applied to hot-spots

fe-safe - identification of hotspots

Identifies the fatigue „hotspots‟

Surface node

Arrow length is proportional to fatigue damage

Can be displayed for whole model or for hot-spots

fe-safe – damage vector plotting

Shows direction of crack initiation

Example from a casting simulation from Magma Abaqus fe-safe

fe-safe – property mapping

Varies the materials properties at each node

Node 1:: Material 1

Node 2 :: Material 2

Mapping model eqn

ID for each property

Node 1:: Material 1

Node 2 :: Material 2

Mapping model eqn

ID for each property

Casting simulation (Finite Difference

model)

Contour plot

of material

properties(Finite Element

model)

A property

mapping file

is generated

fe-safe uses the

mapped nodal

properties for

the fatigue life

analysis

fe-safe – wide choice of algorithms

Strain-life and stress-life

Picture

Automatic detection and analysis of surface contact

Sub-surface crack initiation…..etc

fe-safe can select the algorithm automatically, and

change algorithm in an assembly of different materials

• Network licence, WAN licencing, world-wide licencing

• Distributed processing – validated on a network of over 100

machines

• Multi-processor support – threaded code (no extra costs)

• Batch operation – with parameter modification for parametric

studies

• Save standard analyses as templates

• Can be launched from other applications (Isight, Workbench)

• Windows, Linux, Unix PCs, HP, Sun, IBM-AIX, SGI

• Includes signal processing and fatigue from strain gauges

fe-safe – system features

fe-safe FEA model files -

• Multiple FEA files can be concatenated on input

• More than 1 TerraByte of concatenated files can be input

• No limit to the number of steps and increments (validated for 220

000 steps/increments)

• No limit to the number of load cases that can be superimposed

• Can read integration point, nodal averaged or nodal data

• Can read integration point data and generate nodal averaged or

nodal data

fe-safe – system features

fe-safe plug-in algorithm allows users to write their own fatigue

algorithm

• Located in a dll

• Called for each node

• fe-safe generates the stress tensor sequence

• Supports stress, strain, temperature, time-dependency and

property mapping

• Users may add extra parameters to the materials data base

• Plug-in can be called once to initialise and read in data from

the screen or batch file if required. A GUI is auto-generated.

• Looks exactly like a native fe-safe algorithm

fe-safe – system features

Superimposed load histories

Loading

Single load history

Signal

Summary of Tests - DEF STAN 00-35

0.00001

0.0001

0.001

0.01

0.1

1 10 100 1000 10000

Hz

g2 /H

zPSD

Rainflow

cycles

Modal superimposition – steady

state and random dynamic

+

+

Sequencies of FEA solutions

Complex loading sequences

can be defined easily

Unit load stresses x load histories

superimposed

N repeats

FEA transient

analysis

- non-proportional

N2 repeats

Damage per block

0

5

10

15

20

25

30

35

40

45

1 2 3 4 5 6 7 8

Block

Da

ma

ge

(%

)

Damage for N repeats of each loading block

fe-safe – where does the damage come from?

fe-safe – is a complete system, with few add-ons

Thermo-mechanical fatigue (/TMF)

and creep-fatigue (/TurboLife)

(example – turbine blade)

Fatigue of welded joints using

the Verity method from Battelle.

Includes high temperature fatigue.

The Verity method has been adopted

by the ASME B&PV Design code

Section 8 Div II.

fe-safe – add-ons

fe-safe/composites with Helius:fatigueTM

fe-safe – add-ons

INTERFACES

Input/output

ABAQUS .fil

ABAQUS .odb

NASTRAN f06

NASTRAN op2

ANSYS .rst

I-DEAS .unv

Pro/M s01..., d01

General .csv

Output

Hypermesh .hmres

PATRAN

FEMVIEW

CADFIX

FEMAP

fefe-safe durability analysis from FEA

Redesign

Design

FEAABAQUS, ANSYS

I-DEAS,

NASTRAN, Pro/E

Stress

resultsfe-safe

Loading

Lifecontours

Maximum FEA file size is 1 Terra-byte

Key features

FAST

COMPREHENSIVE

ACCURATE

EASY TO USE

Complex loading

Safety factors

Probability analysis

Multiaxial fatigue

Cast iron analysis

Virtual strain gauges

High temperature fatigue

Welded joints

fe-safe/Rotate

fe-safe/TMF

fe-safe/TurboLife

VerityTM

module in fe-safe

fe-safe/composites with

Helius:fatigueTM

Redesign

Design

FEAABAQUS, ANSYS

I-DEAS,

NASTRAN, Pro/E

Stress

resultsfe-safe

Loading

LifecontoursWidely validated - Widely used

Selected by some hundreds of

leading engineering companies